SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2307.13702 · 2023

Measuring Faithfulness in Chain-of-Thought Reasoning

arXiv · PDF · Open in the Atlas

Code that ran

2 of 3 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 1 violated it and 0 have no contract verdict; 1 failure of our instrument; 1 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2307.13702.json · badge: /api/ran/2307.13702.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
technion-cs-nlp/parametric-faithfulness canonical (classifier label)
classifier basis: classifier_2025
2 ran · 1 no recorded run
FunctionStatusWhere it lives
get_batch_loss Ran
ran, violated the contract we drafted
technion-cs-nlp/parametric-faithfulness/unlearn.py
code served (permissive licence) · get_code("44dd25d964127b38")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned Tensor, ndim 1, dtype family float, shape (2,) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
from torch import nn

torch.manual_seed(0)

# The function expects:
# - output: a tensor that will be sliced [..., :-1, :] and transposed to have shape (..., vocab_size, seq_len)
# - labels: a tensor that will be sliced [..., 1:] and used with CrossEntropyLoss
# 
# CrossEntropyLoss expects input of shape (N, C, d1, d2, ...) and target of shape (N, d1, d2, ...)
# where C is the number of classes (vocab_size)
#
# After output[..., :-1, :] we get shape (batch, seq_len-1, vocab_size)
# After transpose(-1, -2) we get shape (batch, vocab_size, seq_len-1)
# After labels[..., 1:] we get shape (batch, seq_len-1)
# This matches CrossEntropyLoss expectations

batch_size = 2
seq_len = 5
vocab_size = 4

# output shape: (batch_size, seq_len, vocab_size)
# After slicing [..., :-1, :]: (batch_size, seq_len-1, vocab_size)
# After transpose: (batch_size, vocab_size, seq_len-1)
output = torch.randn(batch_size, seq_len, vocab_size)

# labels shape: (batch_size, seq_len)
# After slicing [..., 1:]: (batch_size, seq_len-1)
# Values should be in range [0, vocab_size) or -100 for ignore_index
labels = torch.randint(0, vocab_size, (batch_size, seq_len))

ARGS = {
    "output": output,
    "labels": labels,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 1, dtype family float, shape (2,) — shape (2,) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_linear_schedule_with_warmup Ran
ran; our contract draft was wrong, not the code
technion-cs-nlp/parametric-faithfulness/unlearn.py
code served (permissive licence) · get_code("5bf56e40b4301199")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned LambdaLR. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn
from torch.optim import SGD

torch.manual_seed(0)

# Create a simple model
model = nn.Linear(4, 3)

# Create an optimizer
optimizer = SGD(model.parameters(), lr=0.1)

# Define schedule parameters
num_warmup_steps = 100
num_training_steps = 1000
last_epoch = -1

ARGS = {
    "optimizer": optimizer,
    "num_warmup_steps": num_warmup_steps,
    "num_training_steps": num_training_steps,
    "last_epoch": last_epoch,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type LambdaLR — a LambdaLR output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a LambdaLR output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
run_lm_eval Not yet run
unverified (harvested, no recorded run)
technion-cs-nlp/parametric-faithfulness/unlearn.py
code served (permissive licence) · get_code("2069990162e50256")
How it ran

no invocation recorded.

Also present in other papers' code

Of 3 samples lifted from this paper's linked repositories, 3 carry a function identity and 0 do not. Of the 3 distinct functions with an identity, 2 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 2 later) and 1 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
get_linear_schedule_with_warmup 21 other papers
8 dated earlier, 13 later
2004.10964 (2020-04)
Don't Stop Pretraining: Adapt Language Models to Domains and Tasks
2011.07430 (2020-11)
Audio-Visual Event Recognition through the lens of Adversary
2109.04912 (2021-09)
ReasonBERT: Pre-trained to Reason with Distant Supervision
and 18 more not named here
get_batch_loss 4 other papers
4 later
2409.11844 (2024-09)
MEOW: MEMOry Supervised LLM Unlearning Via Inverted Facts
2502.14829 (2025-02)
Measuring Chain of Thought Faithfulness by Unlearning Reasoning Steps
2506.01339 (2025-06)
Invariance Makes LLM Unlearning Resilient Even to Unanticipated Downstream Fine-Tuning
and 1 more not named here

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Large language models (LLMs) perform better when they produce step-by-step, "Chain-of-Thought" (CoT) reasoning before answering a question, but it is unclear if the stated reasoning is a faithful explanation of the model's actual reasoning (i.e., its process for answering the question). We investigate hypotheses for how CoT reasoning may be unfaithful, by examining how the model predictions change when we intervene on the CoT (e.g., by adding mistakes or paraphrasing it). Models show large variation across tasks in how strongly they condition on the CoT when predicting their answer, sometimes relying heavily on the CoT and other times primarily ignoring it. CoT's performance boost does not seem to come from CoT's added test-time compute alone or from information encoded via the particular phrasing of the CoT. As models become larger and more capable, they produce less faithful reasoning on most tasks we study. Overall, our results suggest that CoT can be faithful if the circumstances such as the model size and task are carefully chosen.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2307.13702")
get_code_for_paper("2307.13702")
have("2307.13702")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2307.13702.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2307.13702.svg)](https://syntology.ai/paper/2307.13702)

Connect an agent — have() is free.